> ## Documentation Index > Fetch the complete documentation index at: https://docs.cloudthinker.io/llms.txt > Use this file to discover all available pages before exploring further. # Recommendations > Look up recommendation attributes, sources, and categories, and move savings opportunities from finding to implemented change The recommendation engine turns detected savings opportunities into prioritized, trackable work items. Each recommendation carries savings, effort, and risk metadata so you can decide what to implement first. ## Recommendation attributes | Attribute | Description | | --------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | | **Title** | Clear, actionable summary of the optimization | | **Description** | Detailed explanation of the issue and solution | | **Potential savings** | Estimated monthly or annual savings | | **Effort level** | Implementation complexity: Low, Medium, or High | | **Risk level** | Potential impact on workloads: Low, Medium, or High | | **Source** | Origin: [CloudKeepers](/guide/infrastructure/cloudkeepers), Conversation, [Assessment](/guide/infrastructure/assessment), or Manual | Recommendations start as drafts, become active when promoted, and are tracked through execution in [Plan](/guide/infrastructure/plan) — the full lifecycle is documented on [Keepers](/guide/infrastructure/cloudkeepers#from-finding-to-recommendation). ## Findings vs recommendations Keeper detections create **findings** first. A finding moves through these statuses: | Status | Meaning | | ---------------- | ---------------------------------------------------------- | | **New** | Just detected; nobody has looked at it yet. | | **Acknowledged** | A team member has seen the finding and owns the follow-up. | | **Active** | Work on the finding is underway. | | **Resolved** | The underlying issue is fixed and verified. | | **Dismissed** | Reviewed and intentionally not acted on. | Promoting a finding creates a **recommendation** — a separate object with its own lifecycle, impact analysis, and implementation playbook. ## Sources | Source | How recommendations are created | | -------------------------------------------------- | ----------------------------------------------------------------------------------------------------- | | [CloudKeepers](/guide/infrastructure/cloudkeepers) | Detection rules continuously scan for utilization drift, cost anomalies, and best-practice violations | | Conversation | Ask [Alex](/guide/agents/alex) with `#recommend` to analyze a specific area | | [Assessment](/guide/infrastructure/assessment) | Well-Architected assessments generate recommendations under the cost optimization pillar | | Manual | Open **Recommendations**, click **New Recommendation**, and set savings, effort, and risk yourself | ## Categories | Category | Typical checks | | ------------------ | --------------------------------------------------------------------------------------------------------------- | | Right-sizing | EC2 instances under 20% average utilization, RDS instances with excess capacity, over-provisioned Lambda memory | | Reserved capacity | 1-year vs 3-year commitment analysis, Savings Plans coverage gaps, reserved instance utilization | | Spot opportunities | Fault-tolerant batch jobs, development/test environments, stateless applications | | Storage lifecycle | S3 Intelligent-Tiering enablement, Glacier archive candidates, infrequently accessed data | | Unused storage | Unattached EBS volumes, orphaned snapshots, empty S3 buckets, unused EFS file systems | | Database | DynamoDB capacity mode, ElastiCache node sizing, index and query optimization, read replica opportunities | | Network | Cross-region transfer, NAT gateway efficiency, VPC endpoints, idle load balancers, cross-zone load balancing | ## Collaboration Each recommendation includes a discussion thread: | Feature | Use it to | | ----------- | ----------------------------------------------- | | Comments | Add context, questions, or implementation notes | | Mentions | Tag team members with `@username` | | Attachments | Link related documents or tickets | | Audit trail | Track all changes and status updates | ## Implementing recommendations **Agent-driven (with approval):** 1. Review the recommendation details. 2. Ask [Alex](/guide/agents/alex) to implement it. 3. Review the proposed changes. 4. Approve execution in the [Approval Center](/guide/approval). 5. Monitor implementation progress. **Manual:** 1. Follow the implementation steps in the recommendation. 2. Use the discussion thread for questions. 3. Record actual savings when done so [Savings Tracking](/guide/cost-optimization/savings) can compare them against projections. ## Exports and integrations | Action | How | | --------- | ---------------------------------------------------------------------------------------- | | Export | Download recommendations as CSV or Excel | | Ticketing | Create Jira tickets from a recommendation | | Sync | Send recommendation updates to external systems via [webhooks](/guide/webhooks/overview) | ## Examples Find and understand recommendations: ```text theme={null} @alex #recommend show the top 10 cost recommendations by savings @alex #recommend list low-effort EC2 recommendations with high savings @alex #recommend explain why we should resize instance i-0abc123 ``` Generate new recommendations for a specific area: ```text theme={null} @alex #recommend analyze our S3 storage costs and suggest optimizations @alex #recommend review EC2 instances idle for more than 7 days ``` Implement one — Alex shows the proposed change and requests approval before executing: ```text theme={null} @alex #recommend implement the resize for instance i-0abc123 ``` ## Related Measure realized savings against projections after implementation Explore the spend trends and anomalies behind recommendations See how detections become findings and promoted recommendations